Analysis of Stepped-Wedge Cluster Randomized Trials: A Tutorial Using Marginal Models

Area of Science:

• Biostatistics
• Clinical Trials Methodology
• Epidemiology

Background:

• Stepped-wedge cluster randomized trials (SW-CRTs) are increasingly used for intervention evaluation.
• Traditional marginal modeling of SW-CRTs often uses suboptimal correlation structures, especially in multi-period designs.
• Recent methodological advancements address these limitations, offering more robust analysis.

Purpose of the Study:

• To survey recent developments in marginal modeling for SW-CRTs.
• To provide practical guidance and case studies for applying these advanced methods.
• To enable researchers to implement paired generalized estimating equations (GEE) and matrix-adjusted estimating equations (MAEE).

Main Methods:

• Focus on multi-parameter within-cluster correlation structures.
• Detailed explanation of paired GEE for simultaneous estimation of mean and correlation parameters.
• Application of matrix-adjusted estimating equations (MAEE) for bias correction in small cluster settings.

Main Results:

• The tutorial surveys methodological developments over the past fifteen years.
• Case studies demonstrate the implementation of GEE/MAEE for SW-CRT analysis.
• The methods are applicable to various SW-CRT designs, including cohorts and repeated cross-sectional samples.

Conclusions:

• Advanced marginal modeling techniques like GEE and MAEE enhance the analysis of SW-CRTs.
• These methods provide more accurate estimation of intervention effects and correlation structures.
• The tutorial empowers applied researchers to utilize these sophisticated statistical tools effectively.